leavelet/cutlass
1
0
Fork 0
Code Issues Pull Requests Actions 25 Packages Projects Releases Wiki Activity

New updates for 2.11 (#775)

Browse Source 
* New updates.

* Minor profiler updates

Co-authored-by: Aniket Shivam <ashivam@nvidia.com>
This commit is contained in:
ANIKET SHIVAM
2023-01-20 16:32:57 -05:00
committed by GitHub
co-authored by Aniket Shivam
parent d49bef88f9
commit 66d9cddc83
1341 changed files with 2003 additions and 1732 deletions
Download Patch File Download Diff File Expand all files Collapse all files
test/unit/gemm/kernel/batched_gemv.cu
+1 -1
View File
@@ -1,5 +1,5 @@
/***************************************************************************************************
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
* SPDX-License-Identifier: BSD-3-Clause
*
* Redistribution and use in source and binary forms, with or without
test/unit/gemm/kernel/testbed_gemv.h
+5 -5
View File
@@ -1,5 +1,5 @@
/***************************************************************************************************
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
* SPDX-License-Identifier: BSD-3-Clause
*
* Redistribution and use in source and binary forms, with or without
@@ -61,7 +61,7 @@ template<typename ThreadBlockShape_,
typename LayoutA_,
typename LayoutB_,
typename LayoutCD_,
int LDG_B = 1, // batch tile size
int THREAD_B = 1, // batch tile size
bool DEBUG=false>
void batched_gemv_kernel_test(cutlass::gemm::BatchedGemmCoord problem_size,
ElementCD_ alpha = ElementCD_(1),
@@ -154,7 +154,7 @@ void batched_gemv_kernel_test(cutlass::gemm::BatchedGemmCoord problem_size,
cutlass::gemm::BatchedGemmCoord tiled_size{ThreadBlockShape::kM,
ThreadBlockShape::kN,
problem_size.k(), // no split-k
DEBUG ? 1 : LDG_B };
DEBUG ? 1 : THREAD_B };
cutlass::gemm::BatchedGemmCoord tiled_shape = swizzle.get_tiled_shape(problem_size, tiled_size);
@@ -334,7 +334,7 @@ template<typename ThreadBlockShape_,
typename LayoutA_,
typename LayoutB_,
typename LayoutCD_,
int LDG_B = 1, // batch tile size
int THREAD_B = 1, // batch tile size
bool DEBUG=false>
void batched_gemv_kernel_perf_test(cutlass::gemm::BatchedGemmCoord problem_size,
ElementCD_ alpha = ElementCD_(1),
@@ -349,7 +349,7 @@ void batched_gemv_kernel_perf_test(cutlass::gemm::BatchedGemmCoord problem_size,
LayoutA_,
LayoutB_,
LayoutCD_,
LDG_B,
THREAD_B,
DEBUG>(problem_size, alpha, beta, true, iter);
}